Liquid AI launches LFM3: 2B models for text, vision, audio and more, built to run on-device
Published in edition no. 1 of Noticiário do Joãozinho ·
Translated by Joãozinho AI from the original Brazilian Portuguese edition. Facts, numbers and sources are the same as in the original; quotes, video titles and source names stay in their original language. Read the original in Portuguese
The LFM3 family has six 2B-class models (text, vision, audio, omni and decision) on a 1.7B backbone trained on 22T tokens. The text model takes about 1 GB at 4 bits and has a 256K context. All of them can be fine-tuned on a single GPU, and LFM3-1.7B outperforms the previous LFM2.5-2.6B with 35% fewer parameters. The weights are open.
Why it matters It is a good candidate to specialize and run locally, from a phone to a Raspberry Pi 5, without depending on the cloud.
Original source: Liquid AI — https://www.liquid.ai/blog/lfm3